As artificial intelligence (AI) and medical imaging continue to transform clinical practice, radiologists-in-training can no longer take a passive role in the march toward this coming change.
This was the position argued by Gerard K. Nguyen, MD, and Anup S. Shetty MD, both with the Mallinckrodt Institute of Radiology at Washington University in St. Louis, in a June 22 Journal of the American College of Radiology article.
“The call for active engagement with AI at the trainee level was made clear as leaders in the field at the RSNA 2017 annual meeting suggested that radiology education in AI may come to be as essential as radiology education in physics,” the pair wrote.
Close the gap in training
Those who are creating this new AI-focused future of radiology aren’t the radiologists themselves, Nguyen and Shetty argued—rather, startup companies are paving the way. But young radiologists don’t need to be experts in these technologies to help shape the future of their profession.
“Lack of advanced programming skills should not be a barrier to active engagement with AI,” authors wrote. “However, familiarity with the language of shared concepts in the field of data science would be a first step.”
For example, when a radiologist uses the term “sensitivity,” an outsider would use the term “recall.” Similarly, “positive predictive value” to imagers equates to “precision” for startups.
Some of these discrepancies in terminology may be easily overcome, some may not, but “the better we can understand and communicate the same language with back end developers, the smoother the integration will be,” Nguyen et al. wrote.
Opportunities to diminish the training gap
Looking within an organization to faculty members and leaders with an inclination toward clinical informatics and AI should be the first step when looking at a training program, the authors urged. The American College of Radiology and the Society for Imaging Informatics in Medicine are resource-rich organizations that can also help.
Additionally, startups look for industry leaders, such as radiology trainees, who have a greater interest in and are more willing to work with technology, which might make for an ideal partnership. These young trainees can ensure organizations have “diverse, reliable and meaningful” data to train models, the pair wrote.
“As the saying goes, garbage in is garbage out, and curating a quality data set requires meticulous attention and proficiency,” authors wrote.
The time has passed for deciding whether the gap needs to be closed, and radiologists need to think about how to close it going forward, they argued.
“We are still in the early phases of AI’s integration with radiology with much work to be done. How we close this gap determines how we can position ourselves to translate these emerging technologies to clinically meaningful change,” Nguyen et al. added.